Tuesday, March 25, 2008

Howto configure a simple TREX index

This will walk you thru the configuration of a simple TREX index on a SAP Netweaver Portal.

This quick guide assumes that you are an administrative user on all the involved systems. First of all, we need to configure the TREX backend-server. This is done by logging into the J2EE administration tool, Visual Administrator.

Go to the Cluster --> Server --> Services --> TREX Service node and select the nameserver.address-field. Enter the correct value for the TREX nameserver. E.g. tcpip://myserver.domain.tld:30001

The correct port can be found by starting the TREXAdmin tool. Here, you will find the info needed on the start-page. Once this value is maintained, click the "Update" button.

Now we are almost ready to create our first search index.

Log on to the Netweaver Portal and follow the steps show here. This will open the "Component Monitor iView" in a new window. Before we continue with the setup, it is important to make sure that all Knowledge Management components and functioning correctly.

In this example case shown below, we get a few common errors, that needs to be corrected - otherwise we might see some strange behaviour of our index.

To correct these specific errors, go to System Administration --> System Configuration --> Knowledge Management --> Content Management --> Global Services --> Scheduler Tasks. Other errors may need some additional post installation steps, check the TREX installation guide or service.sap.com/notes for more info.

Once we have confirmed that all of our KMC components are functioning, we can create our first Index.

From the "Index Administration", create a new Index. The ID must be unique, and if the TREX backend is used for more that one AS Java, I find it useful to include the portal SID. The Name field will be the displayed name of this particular index.

Before we can complete the index creation, a data source must be assigned to this Index. Click the data source link where a list of all KM repositories will be displayed. Select one or more folders and save the Index. The selected repository will now be crawled by the TREX spiders, and all the collected information will be indexed and made searchable.

Please note, that some additional steps will be needed in a productive-like environment, e.g. schedule indexing, permissions, creation of taxonomies and such.

To check the over-all health and status from day to day of your Index(es) and TREX service, open the TREX Monitor iView.

Have fun making unstructured information more accessible ;o)